Hello SFF.NET, I would like to preface this informative article with this: I test equipment to meet certain specifications and standards. I do have a general background in testing, not statistics, calibration, or analysis, so while I do have some expertise, it is not a high level of understanding like an engineer. With this in mind, please feel free to engage me on this subject.
Introduction
As PC building enthusiasts, we often attempt to min-max our builds. We try to get the most out of the hardware we have. One of the highly prevalent topic is the usage of fans (and heatsinks). The purpose of this article is to shed light on how we measure and choose fans.
Now if you want just a ballpark recommendation, this is not for you. This is for the nitty gritty details of how we should conduct real world measurements.
Part I
Critical of Reviews
I have always been critical of the way most reviewers analyze the performance of fans. I find their methodologies questionable, especially when they speak with such authority on their conclusions. But let's get a couple things out of the way first:
1) Good measurement instruments are expensive.
2) Good methodologies are time consuming
3) Good results require good thinking
Good measurement instruments are expensive
Actual accurate, calibrated, and verified instruments are expensive. The standards and applications they follow must be calibrated to ensure that what you are reading is accurate. Manufacturers of calibrated instruments use the NIST traceable system to ensure they meet the standards.
So what does this mean? It means that most tech reviewers out there, youtubers, etc are not using accurate instruments. It may mean that their instrument will not perform consistently. A Fluke calibrated thermometer is $300. That does not include the regular calibration to verify that it is meeting standards!
Still, that is not to say that cheap unverified instruments are worthless. They can give us a general snapshot. Certain type of instruments can be accurate, or verified at home. It just means that the level of accuracy and consistency of test results is lower and should be considered so.
Good methodologies are time consuming
Good testing methods ensure variables are accounted for, the test set up is consistent, and the environment is controlled. This generally takes a lot of time and is beyond the scope of most reviewers. Setting up a quiet chamber to measure performance can also be expensive.
Good results require good thinking
Let's be honest here, most reviewers are not in the business of conducting and evaluating performance. They do not have expertise and background to conduct measurements and any kind of analysis beyond a general impression.
Part II
The Metric to Measure
Now if we're talking about measuring and evaluating the performance of fans, there are a lot of numbers people like to throw at you, saying such and such fan is better than such and such. What do these numbers actually mean?
Decibels (unit of measurement used to express the ratio of one value of a power or field quantity to another on a logarithmic scale)
What people generally measure, when they measure noise, is decibels. But decibels is just a unit, a scale if you will. What most people use is dB SPL, sound pressure level.
Sound pressure level (logarithmic measure of the effective pressure of a sound relative to a reference value, in this case, the threshold of hearing.)
SPL measurements are in essence, the overall energy of the pressure changes that is occurring. It is a sum of all the frequencies being put out. Yet it's highly inaccurate to the way humans perceive sound. That is, lower frequency sound registers more dB SPL, but is heard softer than higher frequency sounds.
Enter equal loudness contour. This is a curve based on research that weighs different frequencies and generate and overall dB SPL that feels closer to how we perceive sound. Lower frequencies are weighed less, and higher frequencies more. Some reviewers are aware of this and use dBA, a type of weighted dB that is closer to perceived hearing.
Yet even then, it is missing a crucial element, a broadband noise feels smoother and more pleasant than a grating squell at a certain frequency, something that an dB reading won't give.
The end all be all: spectrum analysis. If one where to simply describe how spectrum analysis works, one can simply point to the winamp days of audio visualizers. It is simply a dB view of the frequencies recorded. It is not difficult to do a quick spectrum analysis, a high quality microphone run through simple software the shows the levels of the various frequency. It provides an easy for readers to visualize the fan's characteristics. Unfortunately, many reviews lack this feature, and of this writing, I am only aware of Silent PC Review and Cooling Technique who conduct spectrum analysis.
(SPCR's spectrum analysis shows the pitchy nature of this fan)
[TBC]
Introduction
As PC building enthusiasts, we often attempt to min-max our builds. We try to get the most out of the hardware we have. One of the highly prevalent topic is the usage of fans (and heatsinks). The purpose of this article is to shed light on how we measure and choose fans.
Now if you want just a ballpark recommendation, this is not for you. This is for the nitty gritty details of how we should conduct real world measurements.
Part I
Critical of Reviews
I have always been critical of the way most reviewers analyze the performance of fans. I find their methodologies questionable, especially when they speak with such authority on their conclusions. But let's get a couple things out of the way first:
1) Good measurement instruments are expensive.
2) Good methodologies are time consuming
3) Good results require good thinking
Good measurement instruments are expensive
Actual accurate, calibrated, and verified instruments are expensive. The standards and applications they follow must be calibrated to ensure that what you are reading is accurate. Manufacturers of calibrated instruments use the NIST traceable system to ensure they meet the standards.
So what does this mean? It means that most tech reviewers out there, youtubers, etc are not using accurate instruments. It may mean that their instrument will not perform consistently. A Fluke calibrated thermometer is $300. That does not include the regular calibration to verify that it is meeting standards!
Still, that is not to say that cheap unverified instruments are worthless. They can give us a general snapshot. Certain type of instruments can be accurate, or verified at home. It just means that the level of accuracy and consistency of test results is lower and should be considered so.
Good methodologies are time consuming
Good testing methods ensure variables are accounted for, the test set up is consistent, and the environment is controlled. This generally takes a lot of time and is beyond the scope of most reviewers. Setting up a quiet chamber to measure performance can also be expensive.
Good results require good thinking
Let's be honest here, most reviewers are not in the business of conducting and evaluating performance. They do not have expertise and background to conduct measurements and any kind of analysis beyond a general impression.
Part II
The Metric to Measure
Now if we're talking about measuring and evaluating the performance of fans, there are a lot of numbers people like to throw at you, saying such and such fan is better than such and such. What do these numbers actually mean?
Decibels (unit of measurement used to express the ratio of one value of a power or field quantity to another on a logarithmic scale)
What people generally measure, when they measure noise, is decibels. But decibels is just a unit, a scale if you will. What most people use is dB SPL, sound pressure level.
Sound pressure level (logarithmic measure of the effective pressure of a sound relative to a reference value, in this case, the threshold of hearing.)
SPL measurements are in essence, the overall energy of the pressure changes that is occurring. It is a sum of all the frequencies being put out. Yet it's highly inaccurate to the way humans perceive sound. That is, lower frequency sound registers more dB SPL, but is heard softer than higher frequency sounds.
Enter equal loudness contour. This is a curve based on research that weighs different frequencies and generate and overall dB SPL that feels closer to how we perceive sound. Lower frequencies are weighed less, and higher frequencies more. Some reviewers are aware of this and use dBA, a type of weighted dB that is closer to perceived hearing.
Yet even then, it is missing a crucial element, a broadband noise feels smoother and more pleasant than a grating squell at a certain frequency, something that an dB reading won't give.
The end all be all: spectrum analysis. If one where to simply describe how spectrum analysis works, one can simply point to the winamp days of audio visualizers. It is simply a dB view of the frequencies recorded. It is not difficult to do a quick spectrum analysis, a high quality microphone run through simple software the shows the levels of the various frequency. It provides an easy for readers to visualize the fan's characteristics. Unfortunately, many reviews lack this feature, and of this writing, I am only aware of Silent PC Review and Cooling Technique who conduct spectrum analysis.
(SPCR's spectrum analysis shows the pitchy nature of this fan)
[TBC]
Last edited: